×

Announcing: Slashdot Deals - Explore geek apps, games, gadgets and more. (what is this?)

Thank you!

We are sorry to see you leave - Beta is different and we value the time you took to try it out. Before you decide to go, please take a look at some value-adds for Beta and learn more about it. Thank you for reading Slashdot, and for making the site better!

Comments

top

Ask Slashdot: Resources For Kids Who Want To Make Games?

serviscope_minor Re:Good god (112 comments)

Jesus christ, man, he's 11. Get him RPG Maker and let him figure out how to make a game with his own made-up story behind it.

Huh?

I don't think I'm exceptional by the standards of kids who learned to program. I started programming at age 11 because I wanted to write games. I managed to get a simple space invaders type game written myself from scratch in BASIC.

I the kid wants to learn to program games, 11 is an OK age to start.

yesterday
top

Calculus Textbook Author James Stewart Has Died

serviscope_minor Re:What about that stupid book is worth US$244? (166 comments)

I really fucking hate this about academia.

Nope, you mean "American undergraduate university teaching", not academia.

This is not an academic thing. This is something very peculiar to undergraduate teaching in the US.

I think partly it is the obsession with setting millions of questions for students to do. That way one can make it easy on the lecturer by declaring that the student just do a bunch from a textbook.

The system I went through doesn't even remotely work in that way. At the very beginning the lecturer in question (which eventually included me, at least for a while) made a list of recommended books. There were usually about 4 or 5 of unspecified edition, and there would be a bunch of some of them in the various libraries. Students were very much NOT expected to buy any of them unless they really wanted to. As the courses got more specialised, the list of textbooks would get longer (as no book covered everything), then disappeared completely when it became too cutting edge.

Some of the lecturers more advanced in years would occasionally give a glowing recommendation to a book that went out of print some time during the paleocene. I suspect it was a book they found useful as an undergrad and never checked to see if was still in print.

We then lectured. Every so often, a sheet of about 10-15 questions was handed out such that there were about 4 sheets in a 16 lecture course. The questions and lecture notes are generally handed to the next person when the course moves on to a new lecturer for them to use or ignore as they see fit.

It works well. The question sheet means that a specfic edition of a book or even a specific book at all is not required. Us writing the questions means that there's more incentive to have a small number of good questions rather than vast heaps of busy work.

yesterday
top

Calculus Textbook Author James Stewart Has Died

serviscope_minor Re:What are the implications for the textbook mark (166 comments)

I teach out of a thermodynamics text that gets churned every year or so.

Well, then don't. I went through university in the English system, up to and including being a lecturer for a while. The simple solution to this problem is simply to NOT teach out of a text book in this way. It is simply not in the unicersity culture here to to that.

Textbooks are helpful but the students do not need THAT specific textbook.

The first thing to do is write the questions yourself[*]. They're not nearly as hard to write as exam questions because frankly if you screw up a bit on one or two it matters much, much less, they also don't have to be a consistent length or difficulty. You also have a textbook full of questions for insipration. On the courses I was teaching, the lecturers would always hand materials to the next person, and the question sheets often had the year in which the questions were written. One nice undergraduate reminded my of my age by declaring that some of the questions were older than he was.

So, find a few good textbooks and recommend them to the students at the beginning of the course, as a genuine recommendation and not a recommend but I actually mean you have to buy this kind of thing. Then give your lectures and set your questions. The students can then work from the lectures, or any edition of any textbook.

[*] One problem is it seems the American system is based on setting vastly insane numers of questions, which may make this more difficult.

yesterday
top

Calculus Textbook Author James Stewart Has Died

serviscope_minor Re: Math author dies rich... (166 comments)

Really though, the last ethics class I took required an e-book with a 3 use license and six month expiration that cost $130. So, after six months, there is no access to the material at all, like a returned library book without even the value of a paper-bound book that could be burned for warmth.

Well, ironically that probably taught you a lot about ethics.

yesterday
top

Colorado Sued By Neighboring States Over Legal Pot

serviscope_minor Re:Simple answer... (463 comments)

Why does jay-walking turn you into an evil anti-american, versus walking ten feet down the road to use the crosswalk?

Why indeed? Jay-walking is another of those insane laws. We get around fine without jaywalking laws in the UK. It's much more pedestrian friendly for a start.

2 days ago
top

FBI Confirms Open Investigation Into Gamergate

serviscope_minor Re:Most Unbiased Slashdot Gamergate Article (539 comments)

The tobacco industry, big pharma... okay, I'm out.

Well, I was thinking of Hitler and the Nazis. They're always a good standby for extremism :)

2 days ago
top

FBI Confirms Open Investigation Into Gamergate

serviscope_minor Re:Most Unbiased Slashdot Gamergate Article (539 comments)

You should really look into dropping Gamergate entirely, to divest yourself of its now relatively toxic branding, and creating several focused movements to replace it.

"relatively" toxic branding. I like precision. I can indeed think of things with more toxic branding, but not all that many mind you.

2 days ago
top

Marissa Mayer's Reinvention of Yahoo! Stumbles

serviscope_minor Re:Bewitched? (219 comments)

With the obsessive focus on quarterly profits, 2 years is a very long time. In reality, esprically with a company that size it is not. Large organisations are very slow to steer.

2 days ago
top

Investigation: Apple Failing To Protect Chinese Factory Workers

serviscope_minor Re:BBC should tale a good look at itself first (191 comments)

Yep, the BBC is a single homogeneous unit, so the higher ups protecting Saville during his active years 20 years ago or whatever are EXACTLY the same people as those doing the investigation. On that note we should arrest the geniuses at the local Apple store for human rights abuses because they are clearly the same people.

Not to mention their disgraceful one side coverage of the Scottish referendum on Independence this year have left many like myself really not giving much of a shit as to what they have to "report" these days.

Well, Salmond's ludicrous wishlist er, I mean plan for independence was fatally flawed in many ways. The thing is the case against was mostly "the case for is really flawed". Which was true. But yeah, journalists should give equal weight to each sides. Teach the controversy!

2 days ago
top

FBI Confirms Open Investigation Into Gamergate

serviscope_minor Re:cowardice (539 comments)

Just as gamergate is skeptical of conspiracies like 'patriarchy theory'.

I like how instead of responding to the GP and addressing his points, you jump on to a complete non-sequiteur and start beating that to death.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:Clickbait (129 comments)

I called it cheating because they violated both one of the prime rules of AI: train on a data set that is more or less representative of the data set you will test with, and one of the prime rules of statistics

But they're not trying to do that. They're trying to debunk the claims of "near human" performance, which they do very nicely by showing that the algorithms make vast numbers of mistakes when the data in is not very, very close to the original data.

They also present a good way of finding amusing failure cases. I'd never thought of optimizing misclassifications to find how and where an algorithm fails.

2 days ago
top

Marissa Mayer's Reinvention of Yahoo! Stumbles

serviscope_minor Yeah, don't focus on products. (219 comments)

Sounds like a plan.

Products are for suckers.

They should focus on social clouds for wearable augmented reality drones.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:seems a lot like human vision to me (129 comments)

I think I understand... vaguely. To simplify, you're saying it's been trained on a specific dataset, and it chooses whichever image in the dataset the input is most like.

A bit.

It's easier to imagine in 2D. Imagine you have a bunch of height/weigt measurements and a lable telling you whether a person is overweight. Plot them on a graph, and you will see that in one corner people are generally overweight and in another corner, they are not.

If you have a new pair of measurements come along with no label, you could just find the closest height/weight pair and use that. That is in fact a nearest neighbour classifier. It works, except that you need to keep all the original data around.

If you imagine taking 1000 points along the two axes (1,000,000 in total) you could classify each of them according to who is nearest. If you do that you can see that there is more or less a line separating the two groups.

Machine learning is generally the process of finding that line, or an approximation to it somehow.

The DNNs don't find the nearest neighbour explicitly: they just tell you which side of the line a given input is on. They also have a bunch of domain specific knowledge buit in because we know something about the shape of the line, which helps find it. For example, image objects may be scaled up or down in size or distorted in a variety of ways.

Is that about the gist? I'm probably not going to understand things about higher dimensions without a lot of additional information.

The answer is in fact tied into dimensionality. In the 2D example, you can cover the whole space with 1,000,000 points. In 3D to do the same, you need 1,000,000,000. Beyond that the numbers rapidly become completely infeasible.

2 days ago
top

NASA Tests Feasibility of 3D Printing on the Moon and Other Planets

serviscope_minor Re:The biggest problem is fluid dynamics. (58 comments)

Sure, it's an expensive toy - far more than *I* would be willing to pay certainly - but it squirts plastic out of a nozzle to make weak, crude plastic "toys". Arguably useful, especially when you're $4k/pound away from the nearest general store, but not remotely in the same league as the professional-grade printers working in laser-cured resin, sintered titanium, high temperature ceramics, etc.

Stratasys are the single largest 3D printer company and they sell pretty mich exclusively to businesses. I.e. they're selling them to people who do stuff for money, and only that. That makes them "professional grade" by definition.

The other ones you mention are much slower to run and much more expensive to boot. Not to mention that the resolution/strength is overkill for many applications. Part of being a professional is knowing how to make the right trade-offs and select the correct tool for the job.

You also missed out the starch powder printers which are even weaker than the FDM ones. Another professional tool due to the expense.

2 days ago
top

In IT, Beware of Fad Versus Functional

serviscope_minor Re:In IT, remember to wash your hands (153 comments)

Minivans are what happens when you take a car and stretch it into another vehicle.

Nope. The Nissan Serena was about the same size as a saloon car on the ground. It was nothing at all like a stretched car. In fact it looked more like a van adapted to partial passenger use.

Minivans get crap mileage

I looked at a few CUVs online. They get similar mileage to minivans at the penalty of being able to hold fewer passengers in confort and haul less cargo.

and have crap handling.

Neither of them are race cars. My experience driving minivans is that they provide more than adequate handling for safe operation on normal roads when driven at an appropriate speed for the conditions.

Sure if you try to hammer round a tight curve well above the speed limit, you'll look like a fool at a much lower speed for a minivan than for a Bugatti.

If you like handling then nothing apart from a dedicated sports car will be adequate.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:Clickbait (129 comments)

Why was my characterization of their approach "hardly fair"?

You called it cheating.

Someone -- either the researchers or their press people -- decided to hype it as finding a general failing in DNNs (or "AI" as a whole).

It pretty much is. If you input some data far away from the training set you'll wind up at a completely arbitrary point in the decision boundary.

The research is not at all what it is sold as.

The research shows very nicely that the much-hyped deep learning systems are no different in many ways from everything that's come before. They have a few lovely illustrations of things that fool it, some of which are what you'd get if you follow the decision boundary a good way from the data, rather than jumping in at a random point.

I'd say there's not a huge amount novel in the research, but it's certainly not cheating.

Don't multi-class identification networks typically have independent output ANNs, so that several can have high scores?

My understaning is that they usually have one output node per class, but the previous layers are all common to the different classes.

I assumed, perhaps incorrectly, that the 99+% measures they cited were cases where only one output class had a high score, and the rest were low.

I'd expect that too.

If they were effectively using single-class identifiers, either in fact or by considering only the maximum score in a multi-class identifier,

Isn't that uisually how it's done? You have a bunch of outputs the strength of which indicates class/not class for a bunch of classes, then you take max over them to find out which class is dominant. Most ML algorithms are generalised to multiclass by using a one-versus-all or one-versus-one system like that (usually the former since the latter hasa quadratic cost).

Only a relatively few (e.g. trees and therefore forests) naturally support multiple classes.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:Image processing; LIDAR; ADAS perspective (129 comments)

I've done some image processing work.. It seems to me that you can take the output of this Neural network and correlate it with some other image processing routines, like feature detection, feature meteorology, etc;

If you look at the convolutions learned in the bottom layers, you typically end up with a bunch that look awfully like Gabor filters. In other words, it's learning a feature detection stage and already doing that.

Some sort of depth sensing certainly does help.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:Clickbait (129 comments)

The researchers also basically cheated by "training" their distractor images on a fixed neural network.

That's hardly fair: they were trying to find images that fooled the network. What better way to do that than feeding images in until you find a good one (with derivatives).

The only novel finding here is their method for finding images that fool DNNs in practice -- but the chances are overwhelmingly high that a different DNN, trained on the same training set, would not make the same mistake (and perhaps not make any mistake, by assigning a low probability for all classes).

Probably not, but it would stil lclassify the images as something random, probably with high confidence.

and perhaps not make any mistake, by assigning a low probability for all classes

Not likely: there's no good ways yet for these systems to return such information when it it very, far away from s decision boundary. A way of doing that reliably would be a significant breakthrough.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:B-b-b-but Slashdot said...! (129 comments)

I have been assured many, many times by the experts of Slashdot that computers are nowhere near achieving artificial intelligence.

er... and?

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

serviscope_minor Re:So, useless then? (129 comments)

In the early '80s people were laughing about computers trying to play chess.

Were they? I'm not sure they were laughing about it. By the early 90s you could buy rather slick chess computers which had a board with sensors under each square (pressure in the cheap ones, magnetic in the fancy ones), and LEDs up each side to indicate row/column.

You could play them at chess and they'd tell you their moves by flashing the row/column lights. Those weren't just programs by that stage they were full blown integrated consumer products. Of course they would get thrashed by a sufficiently good player then.

A concrete idea of a chess playing computer (people had always imagined such things, the mechanical Turk being a hoax based on such an idea) came up in 1946, when Zuse actually wrote a program for it (untested).

2 days ago

Submissions

top

Ask slashdot: Clusters on the cheap?

serviscope_minor serviscope_minor writes  |  more than 3 years ago

serviscope_minor (664417) writes "Dear Slashdotters,

A friend of mine has recently started a research group. As usual with these things, she is on a shoestring budget and has computational demands. The computational task is very parallel (but implementing it on GPUs is an open research problem and not the topic of research), and very CPU bound.

Can slashdotters advise on a practical way of getting really high bang for buck? The budget is about 4000 GBP (excluding VAT/sales tax), though it is likely that the system will be expanded later.

The computers will probably end up running a boring Linux distro and Sun GridEngine to manage batch processing (with home directories shared over NFS?)."
top

Best removable storage filesystem for Linux?

serviscope_minor serviscope_minor writes  |  more than 4 years ago

serviscope_minor (664417) writes "What filesystem do you use for portable disks, especially large ones, under Linux? FAT is simply not very good. Using a proper filesystem (e.g. ext3) preserves the read/write permissions of the original machine which is rather annoying when the disk is moved to a different machine with differet user IDs. So is there a way to have a good filesystem that supports all the unixy things such as symlinks, and an execute bit, but does not require lots of chown'ing as root when moved to a different machine?"
top

ARM Based netbook.

serviscope_minor serviscope_minor writes  |  more than 5 years ago

serviscope_minor (664417) writes "Shopping in Robert Dyas of all places (note to non English readers, this is a fairly generic hardware store and has only a small selection of electronics at best) I noticed Inkia ARM based netbooks being advertised, though careful readers will note that the specs seem to differ slightly. The specs are the usual netbook ones along with an 800x480 screen 64Meg RAM, 1G flash and a 400 (or maybe 533MHz) Samsung ARM processor and WinCE. So, it looks like the first non-x86 netbooks have arrived. Sadly, this one is rather expensive, being slightly cheaper than the EEE 2G, with a painfully small amount of RAM, less storage and battery power. But this brings up several interesting questions: are they going to get much cheaper, are there ones with more memory, and will it run OpenBSD? The specs are very similar to the Sharp Zaurus 3000 series which runs OpenBSD very well, but running Firefox in 64M is somewhat painful."
top

Liquid explosives: no danger and no plot.

serviscope_minor serviscope_minor writes  |  more than 6 years ago

serviscope_minor (664417) writes "It has already been established in a previous article that bringing down an aircraft with liquid explosives mixed on a pllane would be very difficult. The men accused of the plot werer brought to trial and a verdict has now been reached. There was not enough evidence to convice any of them of targeting a plane. So apparently, there was not much evidence of a plot that could not have worked anyway."
top

Good profiling tools for C/C++ unser un*x?

serviscope_minor serviscope_minor writes  |  more than 7 years ago

serviscope_minor (664417) writes "It should be well known to any developer that you should only optimize parts of a program which need optimizing. And the way to find those parts is through profiling. This simplifies one point: profiling is difficult. The obvious way is to enable profiling in the compiler and use gprof, but this has problems. Firstly there is no point in profiling a program without turning on -O3 (or which ever), since this can change the results dramatically. Secondly, -O3 will inline functions which can ruin profiling results by making them far too coarse. Even if it doesn't do this, there is no way of determining which part of a function is taking up all the time. So that brings me to my question: does anyone know of profiling tools which do not suffer from these problems? My platform is C++ (using g++) on Linux."
top

serviscope_minor serviscope_minor writes  |  more than 7 years ago

serviscope_minor (664417) writes "You heard earlier today that Dell will be shipping Ubuntu on selected models. Naturally, this is interesting to slashdotters. However, the interest generated by a wider audience will ultimately be more important. Well, apparently, this is the 3rd most popuar topic on the BBC at the moment. So apparently this is interesting to a general audience. I believe that this bodes very well for the future."

Journals

Slashdot Login

Need an Account?

Forgot your password?